Contingency Management treatments of substance abuse are highly effective. However, even these effective treatments fail for many individuals. These treatments fail either when hard to treat individuals never emit the target behavior or when individuals with early success emitting the target behavior never develop long strings of the target behavior. Two procedures designed to address these problems were developed recently. The first is shaping using percentile schedules. Shaping is designed to develop the target behavior in the hard-to-treat. The second is the use of escalating payments. Escalating payments are designed to differentially reinforce longer sequential strings of target behavior following early success. We are conducting studies in smokers using percentile schedules to shape lower breath carbon monoxide (BCO) levels. These studies show that percentile schedules insure contact with the contingencies and promote better treatment outcomes than traditional fixed abstinence criteria (BCO <4 ppm) in the hard-to-treat. However, the amount earned for their first BCO <4 ppm by hard-to-treat in the percentile condition was greater than the amount earned by those in the fixed criterion group. This incentive magnitude difference rather than shaping may be responsible for the increased efficacy of the percentile treatment. One specific aim of this application is to examine if percentile schedules are still more effective than fixed abstinence criteria in the hard-to-treat when this incentive differential is eliminated. The second treatment element designed to improve outcomes with contingency management treatments is escalating payments. These are now part of most contingency management programs. The benefit of these has been seen in analog studies in smokers with early success, but who were not seeking to quit smoking (e.g., Roll &Higgins, 2002). However, there has been no direct test of whether escalating payments improve contingency management outcomes for early successes in a treatment study. A second specific aim of this application is to examine if escalating payments improve clinical outcomes in smokers with early success who are trying to quit smoking. The evidence that shaping using percentile schedules and the differential reinforcement of longer strings of target behavior using escalating payments improve contingency management outcomes all comes from studies in smokers. Ironically, smoking is the form of substance abuse for which the efficacy of contingency management is least well demonstrated. The third specific aim of this application is to demonstrate that contingency management treatments can increase the initiation and maintenance of abstinence in smokers.